🐛🐛 Bug Report

⚗️ Current Behavior

After checking out to a non-head node, creating a tensor deeper than one level (eg., "x/y/z", "a/b/c/d") throws TensorDoesNotExistError at second level (i.e., Tensor 'y' does not exist for "x/y/z", Tensor 'b' does not exist for "a/b/c/d")

Input Code

import hub

ds = hub.dataset("mem://xyz")
ds.create_tensor("abc")
ds.abc.append(1)
a = ds.commit("first")
ds.checkout(a)
ds.create_tensor("x/y/z")

Error

hub.util.exceptions.TensorDoesNotExistError: "Tensor 'y' does not exist."

Expected behavior/code

Expect tensor to be created.

⚙️ Environment

• Python version: 3.8
• OS: Ubuntu 18.04